[llvm] [AMDGPU][True16][MC] VOP3 profile in True16 format (PR #109031)

[Brox Chen via llvm-commits]

================
@@ -189,49 +189,112 @@
 # GFX11: v_add_lshl_u32 v255, 0xaf123456, vcc_hi, null ; encoding: [0xff,0x00,0x47,0xd6,0xff,0xd6,0xf0,0x01,0x56,0x34,0x12,0xaf]
 0xff,0x00,0x47,0xd6,0xff,0xd6,0xf0,0x01,0x56,0x34,0x12,0xaf
 
-# GFX11: v_add_nc_i16 v5, v1, v2                 ; encoding: [0x05,0x00,0x0d,0xd7,0x01,0x05,0x02,0x00]
-0x05,0x00,0x0d,0xd7,0x01,0x05,0x02,0x00
-
-# GFX11: v_add_nc_i16 v5, v255, v255             ; encoding: [0x05,0x00,0x0d,0xd7,0xff,0xff,0x03,0x00]
-0x05,0x00,0x0d,0xd7,0xff,0xff,0x03,0x00
-
-# GFX11: v_add_nc_i16 v5, s1, s2                 ; encoding: [0x05,0x00,0x0d,0xd7,0x01,0x04,0x00,0x00]
+# W32-REAL16: v_add_nc_i16 v5.l, v1.h, v2.l op_sel:[1,0,0] ; encoding: [0x05,0x08,0x0d,0xd7,0x01,0x05,0x02,0x00]
+# W32-FAKE16: v_add_nc_i16 v5, v1, v2 op_sel:[1,0,0]  ; encoding: [0x05,0x08,0x0d,0xd7,0x01,0x05,0x02,0x00]
+# W64-REAL16: v_add_nc_i16 v5.l, v1.h, v2.l op_sel:[1,0,0] ; encoding: [0x05,0x08,0x0d,0xd7,0x01,0x05,0x02,0x00]
+# W64-FAKE16: v_add_nc_i16 v5, v1, v2 op_sel:[1,0,0]  ; encoding: [0x05,0x08,0x0d,0xd7,0x01,0x05,0x02,0x00]
----------------
broxigarchen wrote:

I will prefer to use the old tag for now

https://github.com/llvm/llvm-project/pull/109031